As machine learning workloads increasingly move to the cloud, bandwidth and storage costs are becoming major contributors to overall deployment expenses. In this study, we examine whether Federated Learning (FL) can reduce these costs while maintaining predictive performance.











